You are on page 1of 11

Engineering Geology 158 (2013) 66–76

Contents lists available at SciVerse ScienceDirect

Engineering Geology
journal homepage: www.elsevier.com/locate/enggeo

Seismic response of embankment dams under near-fault and far-field


ground motion excitation
M. Davoodi b,⁎, M.K. Jafari b, N. Hadiani a
a
Research Assistant in International Institute of Earthquake Engineering and Seismology, Assistant professor, Department of Civil Engineering, Islamshahr Branch, Islamic Azad University,
Islamshar, Iran
b
Dept. of Geotechnical Earthquake Engineering, International Institute of Earthquake Engineering and Seismology, Tehran, Iran

a r t i c l e i n f o a b s t r a c t

Article history: The characteristics of near-fault and far-field earthquakes differ. Records suggest that a near-fault ground
Received 25 June 2012 motion is characterized by a large high-energy pulse and a distinctive pulse shape for the velocity time his-
Received in revised form 8 February 2013 tory. The present study compared near-fault and far-field ground motion effects and investigates the con-
Accepted 17 February 2013
sequence of well-known characteristics of near-fault ground motion on the nonlinear dynamic response
Available online 13 March 2013
of an embankment dam. It was shown that median maximum demands, such as crest settlement and critical
Keywords:
slip surface displacement of the embankment dam, were higher for near-fault ground motion than far-field
Ground motion motion. For pulse-type near-fault input, the maximum demand was a function of spectral shape and the
Near-fault ratio of the pulse period to the fundamental period of the embankment dam. Near-fault ground motion
Far-field with a period pulse 1.5 to 4 times greater than the natural period of the dam created a larger seismic
Embankment dam demand than when the pulse period was equal to or smaller than the natural period of dam. This difference
Seismic response increased up to 1.85 times as the ground motion intensity increased. Velocity-related ground motion inten-
sity indices, such as peak ground velocity, were much more effective than peak ground acceleration to cor-
relate with seismic demand. This was because shear stress level has a direct relation with the velocity time
history of ground motion.
© 2013 Elsevier B.V. All rights reserved.

1. Introduction pulse rather than velocity pulse. The most significant results extracted
from these studies are:
The characteristics of near-fault ground motion and their influence
• Story drift displacement calculated using near-fault ground motion
on civil structures have long interested researchers. Recent significant
input is higher than for far-field ground motion (Chopra and
earthquakes have provided unprecedented research opportunities to
Chintanapakdee, 2001a, 2001b; Kalkan and Kunnath, 2006).
advance the state of knowledge on near-fault issues. Near-fault ground
• The long pulse period for near-fault ground motion makes it more
motion is characterized by a pulse-type wave shape, long pulse period,
severe for tall structures than short ones (Bertero et al., 1978;
abundant long-period components, a high ratio of peak ground velocity
Alavi and Krawinkler, 2004).
to peak ground acceleration and possible significant permanent ground
• Pulse-type ground motion travels through the height of the struc-
displacement.
tures as wave propagation, so the traditional response spectrum
The effects of near-fault ground motion on civil engineering struc-
method may not capture the influence of pulse-type ground motion
tures such as buildings, tunnels, bridges and nuclear power plants
(Iwan, 1997).
have been the subject of recent studies. Most relate the destructive po-
• The methods for predicting the inelastic displacement response of
tential of near-fault ground motion to velocity pulses. Opinions differ,
structures, such as the square root of the sum of squares (SRSS)
however. Hall et al. (1995) believed that the damage potential of
and the sum of absolute values (SAV), may give non-conservative
near-fault ground motion depends on how much ground displacement
results for near-fault ground motion (MacRae and Mattheis, 2000;
occurs during the velocity pulse. Makris and Black (2003) stated that
MacRae et al., 2001).
more attention should be directed to the distinguishable acceleration
Studies on the effect of near-fault ground motion on the seismic
behavior of dams and slope stabilities are limited (Ohmachi and
⁎ Corresponding author.
Kojima, 2003; Bayraktar et al., 2009; Gazetas et al., 2009; Hadiani
E-mail addresses: m-davood@iiees.ac.ir (M. Davoodi), jafari@iiees.ac.ir (M.K. Jafari), et al., 2013) and all agree that the effect of near-fault ground motion
n.hadiani@gmail.com (N. Hadiani). is great. By using one near-fault and one far-field ground motion,

0013-7952/$ – see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.enggeo.2013.02.008
M. Davoodi et al. / Engineering Geology 158 (2013) 66–76 67

X H
A Core B C 2H
Foundation Free Field
H Quiet Boundary

Fig. 1. Maximum cross-section and material zones of Masjed Soleyman embankment dam.

Table 1 definition is not universal because near-fault effects attenuate as


Drained dynamic properties for foundation and embankment materials. distance increases which, in turn, leads to a greater effect of factors
Core Shell Foundation such as magnitude and local site conditions on ground motion. The
distinguishing characteristic of near-fault ground motion is the pulses
A B C
generated by the directivity effect and fling-step effect (Bolt and
Dry unit weight (kg/m3) – 2200 2200 2200 –
Abrahamson, 2003). This pulse-type ground motion often contains
Saturated unit weight (kg/m3) 2200 2350 2350 2350 2400
Poisson's ratio 0.45 0.40 0.40 0.40 0.30 one or more distinct pulses in the acceleration, velocity and displace-
Cohesion (N/m2) 0.0 0.0 0.0 0.0 – ment time histories, most frequently in velocity.
Friction angle (degrees) 30.0 45.0 37.0 45.0 – Directivity effect can be classified into forward, backward and
neutral according to the relative position between the rupture direc-
tion and site location. The propagation of fault rupture toward a site
Bayraktar et al. (2009) showed that near-fault ground motion can at a velocity close to the shear wave velocity causes most of the seis-
create larger displacement and stress in concert dams. Gazetas et al. mic energy from the rupture to arrive in a single large pulse of motion
(2009) used simple pulse-type ground motion as representative of that occurs at the beginning of the record. This phenomenon is called
near-fault ground motion to calculate slope displacement under exci- forward directivity (Somerville, 2003). The radiation pattern of the
tation. He concluded that upper-bound sliding displacement from shear dislocation on the fault causes this large pulse of motion to be
near-fault excitation may substantially exceed the values obtained oriented in the direction perpendicular to the fault plane, causing
from some currently available design charts. Hadiani et al. (2013) in- the strike-normal component of ground motion to be larger than
vestigated the correlation between crest settlement of embankment the strike-parallel component at periods longer than about 0.5 s
dams and different ground motion intensity indices for pulse-like (Somerville, 2003). In some cases, the most severe direction can be
near fault ground motions. By introducing two new ground motion different from the fault-normal direction. For example, maximum
intensity indices, they showed that, inelastic ground motion intensity spectral demand agrees reasonably well with the strike-normal direc-
indices have a higher correlation with the settlement of embankment tion for ground motion recorded within 5 km of strike-slip faults and
dams. can depart significantly from the strike-normal direction for reverse
The aim of the present study was to compare the nonlinear earth- faults (Howard et al., 2005; Davoodi and Hadiani, 2010). On the
quake response of an embankment dam subjected to near-fault and other hand, backward directivity effects that occur when the rupture
far-field ground motion excitations. Masjed Soleiman embankment propagates away from the site give rise to the opposite effect; long
dam, which is 3 km from the Andica fault and has a height of 177 m duration motions of low amplitude at long periods.
(Jafari and Davoodi, 2006), was used to quantify the effect of near-fault Another important aspect of near-fault ground motion is fling, which
ground motion. is the permanent static displacement in the fault-parallel direction for
strike-slip faults or in the fault-normal direction for dip-slip faults.
2. Definition and characteristics of near-fault ground motion Fling is a result of permanent ground displacement that generates one-
sided velocity pulses, whereas forward directivity is a dynamic phenom-
Near-fault ground motion is motion that is typically assumed to be enon that produces no permanent ground displacement and hence
restricted to within 20 km of a fault (Li and Xie, 2007). However, this two-sided (reversing) velocity (Bray and Rodriguez-Marek, 2004).

1 0
Modulus reduction factor

0.8
50
Core
0.6
Height (m)

Core
100 Up-Shell
0.4
Shell Down-Shell
150
0.2
Foundation Foundation
0 200
0.0001 0.001 0.01 0.1 1 10
0 250 500 750 1000
Cyclic strain (%) Shear velocity (m/s)

Fig. 2. Variation of shear modulus of dam materials with strain. Fig. 3. Profile of shear wave velocity in dam body.
68 M. Davoodi et al. / Engineering Geology 158 (2013) 66–76

(a)
Y f (t)

H
X
A Core B C 2H

Fixed Boundary
H Fixed Boundary

(b)

(c)

(d)

Fig. 4. Natural vibration characteristics of embankment dam model: (a) problem set-up for natural frequency determination; (b) applied load; (c) transverse vibration; (d) natural
frequency via FFT power spectrum of transverse vibration.

Both directivity and fling-step can produce large velocity pulses, but associated with the maximum velocity response spectrum. Regression
fling typically involves longer-period motion associated with the lower analysis shows that the period of a near-fault velocity pulse is a function
acceleration spectrum (Graves, 2003). The periods of near-fault velocity of the magnitude of the earthquake and soil-site conditions. This magni-
pulses produced by earthquake magnitudes (Mw) less than 7.5–8 are tude dependence of the pulse period causes the response spectrum to
typically less than 3–4 s. The predominant period of the acceleration have a peak whose period increases with magnitude, such that the
spectrum produced by a strong velocity pulse is approximately 0.75 near-fault ground motions from moderate-magnitude earthquakes
times the pulse period (Somerville, 2003). But Alavi and Krawinkler may exceed those of larger earthquakes at intermediate periods around
(2004) concluded that the pulse period can be measured as the period 1 s (Somerville, 2002; Panza et al., 2011).
M. Davoodi et al. / Engineering Geology 158 (2013) 66–76 69

Table 2a Table 2c
Near-fault ground motion database with short period pulse. Far-field ground motion database.
1
No. Earthquake Station Year Mw ClsD2 Tp (s) No. Earthquake Station Component Year Mw ClsD
(km) (km)

1 Sierra Madre Cogswell Dam 1991 5.61 22.00 0.29 1 Kern County Taft Lincoln School TAF111 1952 7.36 38.89
2 Loma Prieta Gilroy — Gavilan Coll. 1989 6.93 9.96 0.39 2 San Fernando Castaic — Old Ridge Route ORR291 1971 6.61 22.63
3 Parkfield Temblor pre-1969 1966 6.19 15.96 0.40 3 Northridge-01 LA — Century City CC North CCN360 1994 6.69 23.41
4 Loma Prieta Gilroy Array #1 1989 6.93 9.64 0.40 4 Northridge-01 Moorpark — Fire Sta MRP180 1994 6.69 24.76
5 N. Palm Springs Desert Hot Springs 1986 6.06 6.82 0.42 5 N. Palm Hesperia HES002 1986 6.06 72.97
6 Northridge-01 Pacoima Dam (downstr) 1994 6.69 7.01 0.44 Springs
7 Morgan Hill Anderson Dam 1984 6.19 3.26 0.45 6 Whittier Glendora — N Oakbank A-OAK170 1987 5.99 22.11
(downstream) Narrows-01
8 Loma Prieta Gilroy Array #3 1989 6.93 12.82 0.47 7 Coalinga-01 Parkfield — Cholame 2WA H-C02000 1983 6.36 44.72
9 N. Palm Springs Whitewater Trout Farm 1986 6.06 6.04 0.53 8 Loma Prieta Hollister — South & Pine HSP090 1989 6.93 27.93
10 Morgan Hill Coyote Lake Dam (SW Abut) 1984 6.19 0.53 0.75 9 Landers Yermo Fire Station YER360 1992 7.28 23.62
11 Loma Prieta LGPC 1989 6.93 3.88 0.75 10 Whittier Pacoima Kagel Canyon USC A-KAG315 1987 5.99 36.29
12 Whittier Downey — Co Maint Bldg 1987 5.99 20.82 0.81 Narrows-01
Narrows-01 11 Morgan Hill San Juan Bautista, 24 Polk St SJB213 1984 6.19 27.15
13 Morgan Hill Halls Valley 1984 6.19 3.48 0.84 12 Loma Prieta Fremont — Emerson Court FMS090 1989 6.93 39.85
14 Northridge-01 LA — Sepulveda VA Hospital 1994 6.69 8.44 0.85 13 Northridge-01 LA — Saturn St STN110 1994 6.69 27.01
15 Coyote Lake Gilroy Array #6 1979 5.74 3.11 0.91 14 Chalfant Convict Creek A-CVK000 1986 6.19 31.19
16 N. Palm Springs North Palm Springs 1986 6.06 4.04 0.91 Valley-02
17 Gazli, USSR Karakyr 1976 6.80 5.46 1.06 15 Whittier Terminal Island — S Seaside A-SSE252 1987 5.99 40.36
18 Loma Prieta Saratoga — W Valley Coll. 1989 6.93 9.31 1.15 Narrows-01
19 Superstition Hills-02 El Centro Imp. Co. Cent 1987 6.54 18.20 1.50 16 Coalinga-01 Parkfield — Gold Hill 6 W H-PG6000 1983 6.36 47.88
20 Loma Prieta Gilroy Array #4 1989 6.93 14.34 1.50 17 Loma Prieta Hayward — BART Sta HWB310 1989 6.93 54.15
18 Northridge-01 Anaheim — W Ball Rd WBA000 1994 6.69 68.62
1-Mw = earthquake magnitude.
19 Northridge-01 Huntington Bch — Waikiki WAI290 1994 6.69 69.50
2-ClsD = closet distance to fault rupture plane.
20 N. Palm San Jacinto — Valley H06360 1986 6.06 30.97
Springs Cemetery

Chopra and Chintanapakdee (2001a, 2001b) conducted an analyti-


cal investigation of the response of elastic and inelastic single degree-
of-freedom (SDOF) systems to near-fault and far-field earthquake analyses were conducted using the finite difference program FLAC2D
motion. It was concluded that near-fault ground motion has a higher (2005) based on a continuum finite difference discretization using the
peak ground velocity to peak ground acceleration (PGV/PGA) ratio than Langrangian approach. The geometry of the model is shown in Fig. 1.
far-field ground motion, which generates a wider acceleration-sensitive, In order to simulate real boundary conditions, quiet and free-field
displacement-sensitive and much narrower (generally 0.8 b T b 2.5 s)
velocity-sensitive region in the response spectra than far-field motions.
The narrower velocity-sensitive region of near-fault records is shifted to-
wards longer periods and apparently controlled by the dominant velocity
pulse.

3. Dam model and material properties

Nonlinear analyses were performed on a two-dimensional, plain


strain model of Masjed Soleiman embankment dam. The numerical

Table 2b
Near-fault ground motion database with long period pulse.

No. Earthquake Station Year Mw ClsD Tp (s)


(km)

1 Loma Prieta Gilroy Array #2 1989 6.93 11.07 1.56


2 Superstition Parachute Test Site 1987 6.54 0.95 1.90
Hills-02
3 Northridge-01 Sylmar — Olive View Med FF 1994 6.69 5.30 2.58
4 Northridge-01 Jensen Filter Plant 1994 6.69 5.43 2.90
5 Imperial Valley-06 EC Meloland Overpass FF 1979 6.53 0.07 3.00
6 Northridge-01 Sylmar — Converter Sta East 1994 6.69 5.19 3.10
7 Imperial Valley-06 El Centro Array #7 1979 6.53 0.56 3.20
8 Chi-Chi, Taiwan TCU076 1999 7.62 2.76 3.20
9 Imperial Valley-06 El Centro Array #6 1979 6.53 1.35 3.30
10 Nahanni, Canada Site 1 1985 6.76 9.60 3.40
11 Imperial Valley-06 El Centro Differential Array 1979 6.53 5.09 3.70
12 Kocaeli, Turkey Duzce 1999 7.51 15.37 3.80
13 Kocaeli, Turkey Yarimca 1999 7.51 4.83 3.80
14 Chi-Chi, Taiwan TCU129 1999 7.62 1.84 4.10
15 Imperial Valley-06 El Centro Array #8 1979 6.53 3.86 4.20
16 Kocaeli, Turkey Gebze 1999 7.51 10.92 4.30
17 Chi-Chi, Taiwan TCU075 1999 7.62 0.91 4.40
18 Tabas, Iran Tabas 1978 7.35 2.05 4.70
19 Kocaeli, Turkey Sakarya 1999 7.51 3.12 5.70
Fig. 5. (a) Earthquake magnitude and distance range for three record sets, (b) T versus
20 Imperial Valley-06 El Centro Array #10 1979 6.53 6.17 6.10
TPSv.
70 M. Davoodi et al. / Engineering Geology 158 (2013) 66–76

Fig. 6. Acceleration spectrum of records from Tables 2a, 2b and 2c after scaling where PSa(T0) = 0.15 g: (a) benign ground motion; (b) aggressive ground motion; (c) far-field
ground motion; (d) median of each set.

boundaries available in the code were adapted for the horizontal and corresponding to the sharp spikes in the amplitude of the FFT power
lateral boundaries, respectively. The foundation and embankment spectrum were read from the frequency–amplitude plot.
soils were modeled as elastic and elasto-plastic, respectively. Both of These frequencies were interpreted to represent the natural
these constitutive models were combined with Hardin/Drnevich hys- frequencies of vibrations of the problem. The lowest frequency
teretic damping to provide energy dissipation in the elastic range. value corresponded to the first natural frequency (fundamental
The dynamic properties of materials used in the analyses are frequency) of the problem analyzed. Fig. 4 shows the procedures
presented in Table 1. The laboratory modulus reduction factor (G/Gmax) used to determine natural frequencies. The estimated values for the
results for different soil types used in Masjed Soleiman embankment first two resonance frequencies were 0.91 and 1.24 Hz. The results
dam are shown in Fig. 2. A profile of shear wave velocity in the dam of a previous study by Jafari and Davoodi (2006) on this dam using
body is presented in Fig. 3. For an accurate representation of wave trans- in-situ dynamic tests confirm the calculated natural frequencies in
mission through the soil model, the spatial element size, ΔL, must be the present study. It is evident that the frequencies shown in Fig. 4
smaller than approximately one-tenth of the wavelength, λ, associated refer to modal frequencies in the elastic range. These frequencies
with the highest frequency component of the input wave. Based upon constantly change in a nonlinear system as component yielding
the elastic properties and the mesh size of the FLAC model, the excitation gradually progresses.
input record was filtered to remove frequencies above 7 Hz before being
applied to the model. 5. Ground motion data base

4. Determining natural frequency of embankment dam The ground motion database compiled for nonlinear analyses con-
stituted a representative number of far-fault and near-fault ground
Determining the natural frequencies of vibrations is a prerequisite motion. A total of 60 records were selected to cover a range of fre-
to studying time-dependent behavior in a system subjected to an ex- quency contents, durations, amplitudes and distances. They were di-
ternal vibrating stimulus. A full-cycle sinus pulse has a wide range of vided into three subsets, each containing 20 records. The first set
frequencies. A structure subjected to a full-cycle sine pulse excitation contained ordinary far-fault ground motion without pulse character-
will experience a large response whenever one of the frequencies in istics. The next two sets were pulse type near-fault ground motion
the applied excitation approaches one of the structure's natural fre- containing a forward directivity effect with different pulse periods.
quencies of vibration (Chugh, 2007). For determination of the natural The near-fault records with pulse periods 1.5 times shorter than
frequencies of vibration, the excitation function showed in Eq. (1) the natural period of the dam were labeled “benign”. The records
was applied at the crest in the discretized model of the problem in with pulse periods 1.5 times longer than the natural period of the
the x-direction. dam were labeled “aggressive”. All processed ground motions came
from the Next Generation Attenuation Project database (PEER NGA
f ðtÞ ¼ p0 sinð2πf 0 tÞ for 0≤tbt0 Database, 2005) and the near-fault ground motions were oriented
ð1Þ
f ðtÞ ¼ 0 for t > t0 where t0 ¼ 1=f 0 : in the fault-normal direction. Pertinent information on the ground
motion data is presented in Tables 2a, 2b and 2c.
The problem was analyzed for time duration t = dyt and the Fig. 5(a) presents the distribution of the ground motion given in
displacement-time history was saved at a select location. At the end of Tables 2a, 2b and 2c as a function of earthquake magnitude and the
the dynamic analysis, an FFT power spectrum analysis was performed closest distance to rupture. The correlation between the period of
on the displacement-time history data set and the results of frequency pulse (Tp) computed from the velocity traces and the spectral period
versus amplitude of the power spectrum were plotted. Frequencies (TPSv) that corresponds to the peak amplitude of the pseudo velocity
M. Davoodi et al. / Engineering Geology 158 (2013) 66–76 71

spectrum (PSV) is given in Fig. 5(b). The high correlation indicates the
PSa(T0)=0.15 g dependence of spectral quantities on pulse dominant wave forms. For
1
the nonlinear time history (NTH) analyses presented in subsequent
(a) sections, ground motion shown in Tables 2a, 2b and 2c were scaled
to three levels of PSa(T0) (0.15 g, 0.225 g, 0.3 g) and then inputted
0.8 into the embankment dam to compute the seismic response. The
effects of higher-modes, spectral shape and period of velocity pulse
can be investigated using this normalized ground motion intensity
Benign
Relative Height

0.6 method. The pseudo-acceleration spectra of the records after scaling


so that PSa(T0) = 0.15 g and the mean spectrum are presented in
Fig. 6. It can be seen that the aggressive near-fault records do not
Aggresive
show the rapid decrease in PSa for periods exceeding 0.5 s.
0.4

Near Fault 6. Seismic response of dam to near-fault and far-fault ground motion

0.2 In all, 180 nonlinear time history simulations were conducted on the
Far Field embankment dam at three levels of PSa(T0) (0.15 g, 0.225 g. 0.30 g).
Permanent horizontal displacement (PHD) of the dam body and dam
0 crest settlement was used as the primary measure of seismic demand.
-0.2 -0.1 0 0.1
Additional demand measures such as slip surface displacement and
Permanent Horizental Dis (m) acceleration amplification were also investigated. For the PSa(T0), the
median PHD profiles obtained from the NTH analyses of the dam
PSa(T0)=0.225 g subjected to the three sets of ground motions (far-fault, benign ground,
1 and aggressive ground motions) are presented in Fig. 7. As can be seen,
(b) the PHD from aggressive near-fault ground motion excitation is higher
than those resulting from benign and far field ground motions at all
0.8
PSa(T0) levels. This difference can especially be seen between the ag-
gressive and far-field excitation in the upper parts of the dam body.
The mean responses of the upper parts of the dam are approximately
Benign equal for the near-fault (benign and aggressive) and far-fault ground
Relative Height

0.6
motion, but the difference decreases in the lower parts of the dam
body. These results indicate that, unlike benign ground motion, aggres-
Aggresive sive pulse-like ground motion caused relatively severe responses in the
0.4 embankment dam.
The mean acceleration amplification profiles of the embankment
Near Fault center section from the NTH analysis are presented in Fig. 8. Results
0.2 show that the mean acceleration amplification factors induced by
far-field ground motion are larger than the near-fault ones. This phe-
Far Field nomenon for the distribution of acceleration amplification is opposite
to the distribution of permanent horizontal displacement and can be
0
-0.4 -0.2 0 0.2 linked to the higher nonlinearity behavior of the dam during near-
Permanent Horizental Dis (m) fault ground motion.
As the main difference between near-fault and far field ground
motions is highlighted in velocity time history, so, investigation of
PSa(T0)=0.30 g this parameter can be helpful in having an insight into the problem.
1 Previous studies such as Paskaleva et al. (2004) showed the impor-
(c) tance of maximum particle velocity on nonlinear behavior. They
used this parameter to calculate the horizontal strain factor distribu-
tion for Sofia to develop a model that can predict ground failure,
0.8
liquefaction susceptibility assessment and non-linearity deformation
of soil near the surface of this city.
The time-histories of the vertical displacements at the center point of
Benign
Relative Height

0.6 the dam crest obtained by nonlinear analysis using two types of ground
motion are presented in Fig. 9. This figure shows that the settlement
resulting from aggressive near-fault ground motion (Aggressive-5) is
Aggresive higher than that resulting from far-fault ground motion (Far-field-5).
0.4 The largest crest settlement for near-fault shaking is associated with
fewer reversed cycles of loading. This effect is due to the presence of
Near Fault high-amplitude velocity pulses in the aggressive near-fault records
that caused dissipation of sudden energy in a short period of time in a
0.2
single or few excursions. On the other hand, the energy demand on an
Far Field embankment dam subjected to a far-fault motion tends to gradually

0 Fig. 7. Median profile of absolute permanent horizontal relative displacement along


-0.8 -0.6 -0.4 -0.2 0 0.2
center section of dam obtained by NTH analysis: (a) PSa(T0) = 0.15 g; (b) PSa(T0) =
Permanent Horizental Dis (m) 0.225 g, (c) PSa(T0) = 0.30 g.
72 M. Davoodi et al. / Engineering Geology 158 (2013) 66–76

increase over a longer duration, causing an incremental build-up of


PSa(T0)=0.15 g
1 input energy. The significant duration and total input energy variation
for each record are also displayed in Fig. 9.
(a) Significant duration of a ground motion is defined as the length of
the time interval between the accumulation of 5% and 95% of ground
0.8
motion energy. Ground motion energy is defined by the Arias intensi-
ty (Trifunac and Brady, 1975). It can be inferred from Fig. 9 that the
magnitude of the input energy or peak ground acceleration alone is
Relative Height

0.6 not a complete measure of the severity of the ground motion. For ex-
Benign ample, peak demand could be caused by ground motion with less
input energy. In order to interpret the results of Fig. 9, it is important
to consider the definition of input excitation. For the two dimensional
0.4 Aggresive case of SH-waves propagating in the x–y plane, displacement only
occurs in the horizontal direction in x–z plane. The particle velocity
and shear stress in this direction can be calculated as Eqs. (2) and
Near Fault
0.2
(3) respectively:

dw εxz dz σ xz Cs dt σ xz σ σ
Far Field vs ¼ ¼ ¼ ¼ C ¼ xz C ¼ xz ð2Þ
dt dt G dt G s ρC2s s ρCs
0
1 2 3 σ xz ¼ 2ρCs vs : ð3Þ
Accelration Amplification
Here, w = horizontal displacement in the z direction, εxz = shear
PSa(T0)=0.225 g strain, σxz = applied shear stress; ρ = mass density = shear modu-
1
lus, Cs = shear wave velocity; vs = input velocity time history
(b) (input shear particle velocity). It should be noted that, Factor of two
be added to Eq. (2) because 1/2 of the stress is absorbed by the vis-
0.8 cous dashpots in quiet boundaries. Note that although concepts of
Eqs. (2) and (3) are theoretically correct only for linear behavior,
they have been observed empirically to provide a useful insight into
nonlinear behavior such as liquefaction (e.g., Paskaleva et al., 2004).
Relative Height

0.6
Benign Based on Eq. (3), a velocity record can be converted to a stress record
and applied to a quiet boundary. This equation shows that input shear
stress time history has a direct relationship to velocity and no direct
0.4 Aggresive relationship to acceleration or displacement. This phenomenon is
one of the most important keys to aggressive near-fault ground mo-
tion that creates larger inelastic displacement than far-field motion.
Near Fault Generally, aggressive near-fault ground motion has a larger peak
0.2
ground velocity than far-field ground motion, which leads to a greater
shear stress level that can reach the yield surface. In this situation,
Far Field materials may undergo plastic deformation. It is normal for the profile
0 of median acceleration amplification for far-field ground motion to be
1 2 3 greater than the near-fault motion.
Accelration Amplification The peak ground velocity is a useful parameter for the characteriza-
tion of ground-motion intensity index. The velocity is less sensitive to
PSa(T0)=0.30 g the higher-frequency components of the ground motion, so the PGV
1
should be more useful than the PGA in characterizing the damaging po-
(c) tential of ground motion (Panza et al., 2011). To emphasize the impor-
tance of PGV to PGA on the dynamic response of an embankment dam
0.8 in near-fault ground motion, a permanent displacement of the slip
surface of the embankment dam was calculated. The location of the
critical slip surface is shown in Fig. 10(a) and the procedure to calculate
permanent displacement of the slip mass by integration of the average
Relative Height

0.6 acceleration is shown in Fig. 10(b). The variation of the slip surface dis-
Benign
placement (SSD) with PGA and PGV of near-fault ground motion is
presented in Fig. 11. It can be seen that peak ground velocity exhibits
0.4 Aggresive a strong correlation (ρ > 0.8) to SSD whereas PGA shows a weak corre-
lation. It can be concluded that aggressive ground motion having a
greater velocity pulse creates a greater slip surface displacement.
Near Fault The variation of slip surface displacement for different ground mo-
0.2 tions at the second intensity level is presented in Fig. 12. Results of
the linear correlation between the dam response and ground motion
Far Field
0 Fig. 8. Profile of median acceleration amplification along center section of dam
1 2 3 obtained by NTH analysis: (a) PSa(T0) = 0.15 g; (b) PSa(T0) = 0.225 g, (c)
Accelration Amplification PSa(T0) = 0.30 g.
M. Davoodi et al. / Engineering Geology 158 (2013) 66–76 73

Fig. 9. Settlement demands at dam crest: (a) far-field ground motion (Far-field-5); (b) aggressive ground motion (Aggressive-3).

characteristics are summarized in Table 3. Correlation coefficients in- Conversely, the lowest responses were associated with records
creased for velocity-related ground motion as ground motion intensity having Tp/T0 values of approximately 1. In this case, a PSa(T0) value
level rose. The results show that PGV is a better predictor than PGA of will be large because the energy from a pulse with a period of approx-
nonlinear behavior. imately T0 implies that the record is intense as measured by PSa(T0).
However, as the embankment dam begin to behave nonlinearly, its
7. Period of pulse and seismic response of embankment dam period lengthened into a range where there is comparatively less
energy (response spectrum of Benign-18 in Figure 14). Finally, for
Another important parameter of pulse-like near-fault ground mo- Tp/T0 values from 0.2 to 0.3, the pulse excited higher modes of the
tion that can affect the response of an embankment dam is the period embankment dam although PSa(T0) did not detect it (response spec-
of the velocity pulse with respect to the natural period of the embank- trum of Benign-3 in Figure 14).
ment dam. In Fig. 13, crest settlement and slip surface displacement Variations of the median of crest settlement, slip surface displace-
are plotted against the ground motion pulse period for the second in- ment and acceleration amplification factors obtained from 20 records
tensity level. To ascertain the contribution of higher modes and spectral in each set-versus different ground motion intensity levels are presented
shape, it is necessary to inspect both the acceleration and velocity spec- in Fig. 15(a), 15(b) and 15(c), respectively. Aggressive ground motion
tra of the ground motion. Fig. 14 depicts the spectral velocity of the with Tp/T0 values larger than 1.5 did not affect the structure significantly
three types. For Tp/T0 values between 2 and 4, the response was rela- at the low PSa(T0) associated with lower nonlinearity. However, this
tively large compared to the response from records with shorter- type of ground motion created a greater response than other types of
period pulses. This is because the PSa(T0) ground motion intensity motion at high levels of ground motion intensity. Higher-mode effects
index only measures the intensity of the ground motion at T0. Since are seen in Fig. 15 to be important at low PSa(T0) levels. This is clear in
the embankment dam behaved nonlinearly and its effective period Fig. 15(d), which presents variations in crest settlement for each record
lengthened, it was greatly affected by velocity pulses at longer periods at the first intensity level. It can be seen that crest settlement for benign
(response spectrum of Aggressive-5 in Figure 14). records with Tp/T0 b 0.5 (Benign-1 to −10) is as great as that for
74 M. Davoodi et al. / Engineering Geology 158 (2013) 66–76

(a)

(b)

Fig. 10. (a) Critical slip surface for upstream side of dam; (b) integration of acceleration to determine down slope displacement based on Newmark method.

Fig. 11. Variation of slip surface displacement for PSa(T0) = 0.15 g against: (a) PGA; Fig. 12. Variation of slip surface based on Newmark method for different ground mo-
(b) PGV. tions: (a) PSa(T0) = 0.15 g; (b) PSa(T0) = 0.225 g.
M. Davoodi et al. / Engineering Geology 158 (2013) 66–76 75

Table 3
Correlation coefficient between crest settlement and different properties of records (%).

Intensity level Intensity index Correlation coefficient

Near fault Far field

I PGA 60.6 80.3


PGV 64.8 69.9
Arias intensity 87.1 72.0
Cumulative absolute velocity 73.8 81.9
IVA⁎ 90.9 82.2
II PGA 37.5 75.7
PGV 65.7 62.9
Arias intensity 76.7 69.0
Cumulative absolute velocity 82.5 85.2
IVA 84.3 78.8
III PGA 18.3 70.0
PGV 66.2 59.6 Fig. 14. Pseudo velocity spectrum for different types of ground motion after scaling
Arias intensity 61.1 63.3 where PSa(T0) = 0.225 g.
Cumulative absolute velocity 80.7 83.6
IVA 73.6 73.2

IVA = the square root of the product of peak ground velocity and plain integral of the
Spectral amplification factors and nonlinearity behavior are a
squared acceleration (Hadiani et al., 2013). function of earthquake intensity. As shown in Fig. 15(c), acceleration
amplification factors decreased as nonlinearity behavior increased.
Consequently, amplification factors were larger for linear behavior
aggressive and far field ground motions. The median crest settlements than for nonlinear behavior. It can also be seen that aggressive ground
for the aggressive ground motion are up to 1.85 and 1.25 times greater motion had small amplification factors compared to far-field motion
than the benign and far-field ground motions, respectively. These because of their higher nonlinear behavior.
values change to 2.35 and 2.1, respectively, for surface displacement.
However, the median crest settlement for near-fault ground motion 8. Conclusion
(including benign and aggressive) is approximately equal to those of
the far-field motion. The comparison of near-fault and far-field ground motion effects on
the nonlinear dynamic behavior of a high embankment dam was studied
by carrying out 180 non-linear time history analyses at three intensity
levels. The embankment dam was subjected to large displacement de-
mands by the arrival of velocity pulses of near-fault ground motion. Con-
sequently, the dam was required to dissipate considerable input energy
in a single, or relatively few, cycles. In contrast, far-field ground motion
applied input energy more gradually to the dam body over more cycles.
The numerical simulations carried out in this study show that:
• Velocity-related ground motion intensity indices, such as peak
ground velocity and cumulative absolute velocity, are much more
effective than acceleration ground motion intensity indices, such
as peak ground acceleration, to correlate with seismic demand at
high intensity level and nonlinear behavior.
• Tp/T0 and spectral shape play important roles in the seismic response
of embankment dams. Based on this analysis, near-fault records with
Tp/T0 > 1.5 (aggressive) tend to cause a larger seismic response than
far-field or near-fault ground motion with Tp/T0 b 1.5 (benign).
• The effects of aggressive-type ground motion become more impor-
tant as ground motion intensity and nonlinear material behavior
increase.
• The median crest settlement for aggressive ground motion is up to
1.85 and 1.25 times greater than for benign and far-field ground mo-
tions, respectively. These values increase to 2.35 and 2.1, respectively,
for surface displacement. However, median crest settlement for the
near fault ground motion (benign and aggressive) is approximately
equal to that for the far field motion.
• Aggressive ground motion has a small amplification factor compared
to the far-field motion because of its higher nonlinear behavior.
This study was based on a comprehensive dynamic analysis of a
high embankment dam. To be able to generalize the results, such
analyses should be carried out on different embankment dams of var-
ious heights and with different dynamic properties.

Acknowledgments

Fig. 13. Variation of seismic response against Tp/T0: (a) crest settlement; (b) slip sur- This paper is a part of the research project with the title of “Effect of
face displacement for PSa(T0) = 0.225 g. Near Field Ground Motion on Seismic Behavior of Embankment Dams”
76 M. Davoodi et al. / Engineering Geology 158 (2013) 66–76

Fig. 15. Dam response against ground motion intensity: (a) crest settlement; (b) slip surface displacement; (c) acceleration amplification factor; (d) crest settlement for types of
ground motion at first intensity level.

funded by the International Institute of Earthquake Engineering and Hadiani, N., Davoodi, M., Jafari, M.K., 2013. Correlation between settlement of embank-
ment dams and ground motion intensity indices of pulse-like records. Iranian
Seismology (IIEES) under the activity code of 6129 and the project Journal of Science and Technology Transactions of Civil Engineering 37 (1), 111–126.
code of 461. This support is gratefully acknowledged. Hall, J.F., Heaton, T.H., Halling, M.W., Wald, D.J., 1995. Near-source ground motion and
its effects on flexible buildings. Earthquake Spectra 11, 569–605.
Howard, J.K., Tracy, C.A., Burns, R.G., 2005. Comparing observed and predicted directivity
in near-source ground motion. Earthquake Spectra 21, 1063–1092.
References Iwan, W.D., 1997. Drift spectrum: measure of demand for earthquake ground motions.
Journal of Structural Engineering 123 (4), 397–404.
Alavi, B., Krawinkler, H., 2004. Behaviour of moment resisting frame structures Jafari, M.K., Davoodi, M., 2006. Dynamic characteristics evaluation of Masjed Soleiman
subjected to near-fault ground motions. Earthquake Engineering and Structural dam using in-situ dynamic tests. Canadian Geotechnical Journal 43 (10), 997–1014.
Dynamics 33, 687–706. Kalkan, E., Kunnath, S.K., 2006. Effects of fling-step and forward-rupture directivity on
Bayraktar, A., Altunişik, A., Sevim, B., Kartal, M., Türker, T., Bilici, Y., 2009. Comparison the seismic response of buildings. Earthquake Spectra 22 (2), 367–390.
of near- and far-fault ground motion effect on the nonlinear response of dam– Li, S., Xie, L.L., 2007. Progress and trend on near-field problems in civil engineering.
reservoir–foundation systems. Nonlinear Dynamics 58, 655–673. Acta Seismologica Sinica 20 (1), 105–114.
Bertero, V.W., Mahin, S.A., Herrera, R.A., 1978. A seismic design implications of near- MacRae, G.A., Mattheis, J., 2000. Three-dimensional steel building response to near-
fault San Fernando earthquake records. Earthquake Engineering and Structural fault motions. Journal of Structural Engineering 126 (1), 117–126.
Dynamics 6 (1), 31–42. MacRae, G.A., Morrow, D.V., Roeder, C.W., 2001. Near-fault ground motion effects on
Bolt, B.A., Abrahamson, N.A., 2003. Estimation of strong seismic ground motions. In: Lee, simple structures. Journal of Structural Engineering 127 (9), 996–1004.
W.H.K., Kanamori, H., Jennings, P.C., Kisslinger, C. (Eds.), International Handbook Makris, N., Black, C., 2003. Dimensional analysis of inelastic structures subjected to
of Earthquake and Engineering Seismology, vol. 81B. Academic, IASPEI, San Diego, near-fault ground motions. Report No. 03-05. California Earthquake Engineering
pp. 983–1001. Research Center, University of California, Berkeley, pp. 89–91.
Bray, J.D., Rodriguez-Marek, A., 2004. Characterization of forward-directivity ground Ohmachi, T., Kojima, N., 2003. Near-field effect of hidden seismic faulting on a concrete
motion in the near-fault region. Soil Dynamics and Earthquake Engineering 11, dam. Natural Disaster Science 25, 7–15.
815–828. Panza, G.F., La Mura, C., Romanelli, F., Vaccari, F., 2011. Earthquakes, strong-ground
Chopra, A.K., Chintanapakdee, C., 2001a. Comparing response of SDF systems to near- motion. Encyclopedia of Solid Earth Geophysics 1, 252–260 (Springer).
fault and far-fault earthquake motions in the context of spectral regions. Earthquake Paskaleva, I., Panza, G.F., Vaccari, F., Ivanov, P., 2004. Deterministic modelling for
Engineering and Structural Dynamics 30, 1769–1789. microzonation of Sofia — an expected earthquake scenario. Acta Geodaetica Et
Chopra, A.K., Chintanapakdee, C., 2001b. Drift spectrum versus modal analysis of struc- Geophysica Hungarica 39 (2), 275–295.
tural response to near-fault ground motions. Earthquake Spectra 17 (2), 221–234. PEER NGA Database. Pacific Earthquake Engineering Research Center, University of
Chugh, A., 2007. Natural vibration characteristics of gravity structures. Numerical California at Berkeley.
Methods and Analytical Methods in Geomechanics 31 (4), 607–648. Somerville, P.G., 2002. Characterizing near fault ground motion for the design and
Davoodi, M., Hadiani, N., 2010. Orientation of maximum Newmark sliding block displace- evaluation of bridges. In: Nimis, R., Bruneau, M. (Eds.), Proceedings of the Third
ment in the near-fault region for seismic hazard assessment. Presented at: 4th Inter- National Seismic Conference and Workshop on Bridges and Highways. MCEER,
national Conference on Geotechnical Engineering and Soil Mechanics (Tehran). Buttalo, pp. 137–148.
Gazetas, G., Garini, E., Anastasopoulos, I., Georgarakos, T., 2009. Effects of near-fault Somerville, P.G., 2003. Magnitude scaling of the near-fault rupture directivity pulse.
ground shaking on sliding systems. Journal of Geotechnical and Geoenvironmental Physics of the Earth and Planetary Interiors 137 (1), 12.
Engineering 135 (12), 1906–1921. Trifunac, M.D., Brady, A.G., 1975. A study on the duration of earthquake strong motion.
Graves, R., 2003. Report on evaluation of record data processing for static displace- Bulletin of the Seismological Society of America 65, 581–626.
ment. Working Group 1B Report, Next Generation of Attenuation Formulas. FLAC User Guide, 2005, 3rd Ed. FLAC Version 5.0, Itasca Consulting Group Inc.

You might also like